Inhabited subsurface wet smectites in the hyperarid core of the Atacama Desert as an analog for the search for life on Mars

The modern Martian surface is unlikely to be habitable due to its extreme aridity among other environmental factors. This is the reason why the hyperarid core of the Atacama Desert has been studied as an analog for the habitability of Mars for more than 50 years. Here we report a layer enriched in smectites located just 30 cm below the surface of the hyperarid core of the Atacama. We discovered the clay-rich layer to be wet (a phenomenon never observed before in this region), keeping a high and constant relative humidity of 78% (aw 0.780), and completely isolated from the changing and extremely dry subaerial conditions characteristic of the Atacama. The smectite-rich layer is inhabited by at least 30 halophilic species of metabolically active bacteria and archaea, unveiling a previously unreported habitat for microbial life under the surface of the driest place on Earth. The discovery of a diverse microbial community in smectite-rich subsurface layers in the hyperarid core of the Atacama, and the collection of biosignatures we have identified within the clays, suggest that similar shallow clay deposits on Mars may contain biosignatures easily reachable by current rovers and landers.

The modern Martian surface is unlikely to be habitable due to its extreme aridity among other environmental factors. This is the reason why the hyperarid core of the Atacama Desert has been studied as an analog for the habitability of Mars for more than 50 years. Here we report a layer enriched in smectites located just 30 cm below the surface of the hyperarid core of the Atacama. We discovered the clay-rich layer to be wet (a phenomenon never observed before in this region), keeping a high and constant relative humidity of 78% (a w 0.780), and completely isolated from the changing and extremely dry subaerial conditions characteristic of the Atacama. The smectite-rich layer is inhabited by at least 30 halophilic species of metabolically active bacteria and archaea, unveiling a previously unreported habitat for microbial life under the surface of the driest place on Earth. The discovery of a diverse microbial community in smectite-rich subsurface layers in the hyperarid core of the Atacama, and the collection of biosignatures we have identified within the clays, suggest that similar shallow clay deposits on Mars may contain biosignatures easily reachable by current rovers and landers.
Three rovers will land on Mars in the next years (Perseverance, Rosalind Franklin and Tianwen-1) with two of them (Perseverance and Rosalind Franklin) having the primary goal of seeking preserved biosignatures in clays 1 . NASA's Mars2020 Perseverance rover will land in Jezero crater 2 , an impact crater with a diameter of about 45 km located at the edge of the Isidis Basin on Mars. Orbital images suggest Jezero was the site of an open-basin lake ~ 4 Gyr ago during the Noachian era [3][4][5] . The Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) onboard NASA's Mars Reconnaissance Orbiter identified iron/magnesium-smectites and magnesium-rich carbonates within the Jezero crater fan deposits 6,7 thought to have been transported from the surrounding Nili Fossae region 8,9 (smectites are a well-known group of phyllosilicates, or clay minerals, part of a class of "swelling clays" that can readily exchange H2O and cations in their structures 10 ). ESA's Rosalind Franklin rover will land on Oxia Planum in 2023 11 , a Noachian terrain located on the southwest margin of Arabia Terra 12 . The smectite-rich unit at Oxia is representative of a more widespread aqueously altered unit, exposures of which are found scattered over distances as large as 1900 km and which include the Mawrth Vallis clay-rich region 13 . In addition, NASA's Curiosity rover, which landed in Gale crater in August 2012, has already investigated an Fe/ Mg smectite-bearing unit that was first identified from orbit 14,15 .
The reason for sending rovers to smectite-bearing Martian terrains is two-fold: (1) smectite forms from water-rock interactions, and liquid water is a prerequisite for life as we know it on Earth 16,17 , and (2) swelling clays may enable enhanced preservation of biosignatures 18 . On Mars, phyllosilicates are important secondary

Results and discussions
Our first observation in the studied area took place in the second week of March 2017, while sampling soils in a pit dug in the middle of Yungay (24º5′5.28″S, 69º54′54.25″W) (Fig. 1a,b). Starting 30 cm below the thoroughly characterized halite, gypsum, albite and quartz-rich soil surface and shallow subsurface [36][37][38] , we observed a distinct layer of wet clay mineral-rich soil (Fig. 1c). These wet sediments were also observed in the same pit later in March, July and August 2018. Four additional pits dug in August 2018 in a 4 km radius around the first pit also unveiled wet clay-rich soils in the subsurface (Fig. 1d), confirming the existence of a widespread and sustained phenomenon of sub-surface water availability at Yungay. To the best of our knowledge, this represents the first detection of wet subsurface clay-rich layers (or any other type of subsurface wet minerals) at Yungay or other sites of the hyperarid core of the Atacama.
Dual temperature/relative humidity loggers buried in the soil profile (at the surface, 20 cm and 40 cm below the surface, this last sensor set in the middle of the clay layer) in the first inspected pit in 2018 and 2019 showed that while atmospheric relative humidity widely fluctuated daily (as it usually does in the hyperarid core of the Atacama) 32,42 , relative humidity deeper in the soil profile always remained high and constant over time (Fig. 3a).  (a) From bottom to top, XRD patterns of an oriented clay mount following a step-procedure to identify clay minerals: air-dried, glycolated and heated at 400 and 500 °C. The presence of a broad peak at 12.2 Å which migrated at 16.7 Å in the glycolated pattern and the presence of reflections assigned to illite (I) point to an interstratified illite-smectite specimen. The perpetuation of a sharp peak at 10 Å in all patterns indicates discrete illite is present. The sample also contains chlorite and quartz (Ch and Q). (b) Near-infrared spectra of Yungay clay-size separates. From bottom to top, near-infrared spectra of Yungay clay-size separate ("Yungay"), Al-smectite (montmorillonite, from CRISM spectra library) and illite (from PDS Geoscience spectra library). The Yungay clay spectrum shown is an average of three independent measurements (64 scans, 2 cm −1 of resolution), normalized by the maximum intensity. Dotted lines mark the vibrational bands of the clay minerals: 1. 41 Figure S2, panel a). Relative humidity at a depth of 20 cm showed a minor negative correlation with atmospheric surface temperatures (Supplemental Figure S2, panel b), while within the clay-rich layer at a depth of 40 cm, relative humidity was almost completely uncoupled from the harsh and changing surface conditions (Supplemental Figure S2, panel c).
Below the clay-rich layer, we found another layer composed of granular material (a w 0.762), mainly containing albite (calcian), quartz and orthoclase; and, below this, a hard, rocky stratum (a w 0.750) composed of albite (calcian), quartz and ferrous magnesiohornblende (Fig. 1a). The minor decrease in water activity of the layers below the clay-rich layer shows that the illite-smectite in the clay-rich layer preferentially holds water, but may also provide a source of humidity for layers below it. Standard gravimetric analyses 43,44 of the illite-smectitebearing samples showed that these clays hold up to 20% of their weight in water available for microbial life. The source of water for the wet subsurface clay-rich layer is likely related to unusually intense rain events in Yungay over the past decade, particularly in March 2015 when the region received ~ 35 mm of precipitation over the course of several days 45 . Later rain events even created small lagoons in Yungay in 2017, a phenomenon never observed before in this region in the past 500 years 46 .
Although the exact environmental conditions on early Mars are largely unknown, similarities in mineralogy between soils at Yungay and Noachian-and Hesperian-aged Martian terrains indicate similar conditions to those in Yungay soils may have existed on early Mars. Sulfates and chlorides have been detected from orbit and in situ by rovers and are generally indicative of arid and evaporative environments 47,48 . Many of these salts are located in depressions or impact craters and can be associated with clay minerals 49,50 . Saline waters may have been sourced from groundwater upwelling and/or surface runoff, but the identification of river and delta deposits in some of these craters suggests surface runoff was a factor in these locations, and lake level sequences in Gale crater indicate multiple drying and rewetting cycles 51 . Arid, evaporative environments, indicated by sulfates and chlorides, and local flooding from surface runoff, perhaps from melting snow and ice in the case of early Mars, followed by periods of aridity likely existed on early Mars as they do now at Yungay. www.nature.com/scientificreports/ Clay-rich layer samples showed a total organic content of 0.10% dry weight (dw) (a value ten times higher than the highest organic surface carbon values measured in Yungay) 44,45 , and total nitrogen of 0.01% dw, resulting in a C/N ratio of 10, a value in the higher limit range of other reports of desert microbial biomass 46 . Stable isotopic ratios were − 23.4 ‰ (δ 13 C) and 2.3 ‰ (δ 15 N). Depleted bulk δ 13 C ratios suggest that primary carbon fixation in these wet clay-rich layers may be explained either by a subsurface acetyl-CoA pathway (bacteria and/or archaea) and/or by the downward transport of Calvin cycle products created at the surface, as endolithic metabolically active cyanobacteria have been reported inside halites of the surface of the sampled area 38,40,52 .
Although the heterogeneous distribution of cells in small colonies in the clay samples analyzed complicate the assessment of microbial colonization in this new habitat, we have estimated population levels of at least 9.3 × 10 3 cells per gram of sample using direct counting after SYBR Green staining. Fluorescence microscopy analysis of clay-rich samples from 40 cm depth stained with SYBR Green also revealed that microbial cells in these clay minerals form discrete, isolated colonies encapsulated by an envelope of exopolysaccharides similar to biofilms ( Fig. 4a-d), suggesting a functional collaboration among these cells 53 . In addition, CTC (5-Cyano-2,3-di-p-toly tetrazolium chloride, a redox dye commonly used as cellular indicator of respiratory activity) staining of these same samples clearly showed that these cells are actively respiring, demonstrating that these colonies are metabolically active (Fig. 4e,f), and not in a state of metabolic stasis, as it has been proposed in general for microbial species found in subsurface soils of the hyperarid core of the Atacama 54 .
We used culture-dependent and culture-independent methods to investigate the habitability of the wet clayrich soils identified at Yungay. Direct DNA extraction and the subsequent massive parallel sequencing of 16S ribosomal RNA gene amplicons showed that the majority of the sequences obtained are phylogenetically close to 30 bacterial and archaeal species (Fig. 5a), thus unveiling a microbial diversity higher than that reported for soils of this area 30,40,55 . We found that the highest diversity of species was among the Actinobacteria (i.e., Kocuria, Arthrobacter and other uncultured species of unknown affiliation), followed by halophilic archaea (Halostella, Natromonas, Halorussus, Natrinema), beta-(Undibacterium, Aquabacterium, Massilia and Ralstonia) and gamma-proteobacteria (Acinetobacter johnsonii, Halomonas gudaonensis, Marinimicrobium locisalis), with a few species of Firmicutes (Anaeroccocus, Lactobacillus, Tepidibacter, Streptococcus), cyanobacteria (Halothece) and Bacteroidetes (Salinibacter). Some of these species have been already reported in Yungay soils and other regions of the Atacama, such as Halothece 56 , Corynebacterium 57,58 , Kocuria 59 , Lactobacillus and Streptococcus 60,61 , Anaerococcus 62 , Halomonas and Salinibacter 38,58,62 . Interestingly, we recently reported Halomonas gudaonensis, Marinimicrobium locisalis and Acinetobacter johnsonii as three of only four microbial species able to reproduce in the extremely rare temporary lagoons that formed after the very unusual rains in Yungay in 2017 46 . This suggests that both ecosystems (subsurface clay soils and surface lagoons) were temporarily connected (temporarily, as these lagoons had completely evaporated at the time of our latest pits dug on 2018), thus suggesting that these three microbial species may have originated from the subsurface wet clay soils reported here. The analyses of microbial species in these lagoons showed that high rainfall events annihilated most surface species in Yungay 46 , however, the diversity of microorganisms in the wet clay-rich soils suggests that the subsurface microbial community in Yungay thrives or at least remained unaffected following high rainfall events. Cultivation of clay-rich samples in different growing media allowed the growth of 23 novel halotolerant bacterial isolates (Fig. 5b), with many species phylogenetically close to genera not detected by direct DNA extraction, such as Oceanobacillus, Lysinibacilus, Virgibacillus, Halobacillus and Bacillus. Similar to the species detected by direct DNA extraction from clay-rich samples, the microbial species found by cultivation also match the halotolerant/halophilic species reported by other authors in Yungay, like Halobacillus 63 ; and also other surface and subsurface soils in the Atacama (Halobacillus, Virgibacillus, Oceanobacillus, and Bacillus 64 . The case of Bacillus spp. is worth mentioning, as species of this genus are among the first and one of the most common type of bacteria reported in Yungay 27,65,66 and also in other sites of the hyperarid core of the Atacama [67][68][69] . With the aim of helping inform decisions for the upcoming inspection of Martian smectite by NASA and ESA rovers, we also analyzed what type of biosignatures could be detected in the inhabited wet subsurface smectites of Yungay ( Fig. 6a-c). The lipid analysis from clay-rich samples confirmed the presence of microbial biosignatures (heptadecane and phytol derivates) typically associated with different microorganisms such as sulfate-reducing bacteria (iso/anteiso carboxylic acids), archaea (squalene and crocetane), green non-sulphur bacteria and interestingly, also, cyanobacteria (Fig. 6a,c). Thus, these results not only unveiled the biosignatures produced by the microbial species that inhabit the subsurface clay minerals reported here, but also the molecular remains produced by the microbial species reported in the halites present on the surface 40,70,71 . This is the case of Halothece (also detected through Next Generation Sequencing; NGS), of particular interest because it is the only cyanobacteria reported inside halites on the surface soils of Yungay 71 (the presence of cyanobacteria in the inspected clay-rich samples is ruled out as light does not penetrate 40 cm below the surface, see Supplemental Table S1). The detection of biosignatures of this cyanobacterium on the clay layer lends additional support to the hypothesis that primary carbon fixation in this community can be at least partially explained by the downward seepage of Calvin cycle products synthetized at the surface.
Iso/anteiso pairs of the C 15 and C 17 n-carboxylic acids were relatively abundant in the acidic fraction of the analyzed clay-rich samples (Fig. 6a), consistent with the species reported by NGS sequencing. These are lipidic components characteristic of sulphate-reducing bacteria membranes 72,73 also found in heterotrophic bacteria such as Thermus, Deinococcus or Bacillus 74 . The detection of monounsaturated carboxylic acids such as 16:1ω7; 16:1ω9 or 18:1ω9 are also related to the presence of phototrophic bacteria [74][75][76] . The dominance of the C17 chain within the low molecular-weight n-alkanes (Fig. 6c) and the presence of branched (mostly mono-methylated) n-alkanes supported the presence of cyanobacterial sources too 77-80 , as previously described in other surface sites of the Atacama 46,80 . The detection of pristane and phytane also supports the input from surface cyanobacteria, as both isoprenoids are widely viewed as transformation products of phytol 81 . However, other than cyanobacterial origin, pristane can also be sourced in plants or phytoplankton 82 , and phytane can be sourced Scientific Reports | (2020) 10:19183 | https://doi.org/10.1038/s41598-020-76302-z www.nature.com/scientificreports/ in archaeol, the most commonly reported core lipid in archaea 31,83 . The contribution from archaea may also be inferred from the presence of squalene and crocetane (although squalene can also be synthetized by fungi), other isoprenoids characteristically attributed to halophilic 84,85 or methanogenic/methanotrophic 85 archaea, in line with the archaeal species found by NGS sequencing. In addition to the dominant microbial biomarkers, a contribution from eukaryotes was also inferred from the detection of certain terrestrial lipid biomarkers such as stigmastanol 86,87 (Fig. 6b) and odd high molecular-weight n-alkanes (i.e. C27, C29, and C31 88,89 (Fig. 6a,c). Because these are highly recalcitrant biomolecules, able to resist decay even for billions of years 83 , and considering that the samples were collected at 40 cm depth, their presence may be potentially explained by the aerial input of allochthonous material coming from elsewhere in the distant past 30,69,80 . The aerial transport of such recalcitrant biomolecules have been described in the Atacama in places such as the neighboring Salar Grande 80 and in other   40,56,71 . The use of source-specific lipid biosignatures in these samples allowed us to elucidate the structure of the microbial community and the different carbon assimilation pathways operating in the clay-rich microenvironment. As already mentioned, the biomass isotopic ratio of δ 13 C measured in the biomass (i.e., TOC) of the clay-rich samples was − 23.4‰. Assuming a mean δ 13 C of − 8‰ for the fixed atmospheric CO 2 90,91 , the observed δ 13 C value corresponded to a fractionation of ca. 15‰, which is within the range (δ 13 C from − 11 to − 26‰) of those described for microorganisms assimilating CO 2 by the Calvin cycle 92   www.nature.com/scientificreports/ C 17 homologue relative to the other two low molecular-weight n-alkanes. The dominance of C 17 among the low molecular-weight n-alkanes (Fig. 6) is likely related to an input source of cyanobacteria 81,93,94 . δ 13 C compound specific isotope analysis for higher n-alkanes (C 27 and C 29 ) rather reflects a possible eukaryotic input 94 . Thus, the analysis of Yungay clay-rich soils suggests that the distribution of n-alkanes and other biosignatures also reflect a mixed contribution from different types of microorganisms including cyanobacteria that inhabited surface halite and exogenous eukaryotic material. These results highlight the ability of smectite to preserve different types/sources of biosignatures in an environment as extreme as the hyperarid core of the Atacama, with direct implications for the preservation and discovery of biosignatures on Mars 95 . Finally, we compared the analytical capabilities of the Gas Chromatography-Mass Spectrometry (GC-MS) techniques used on the Yungay clay minerals with the analytical instruments on the Mars rovers Curiosity, Rosalind Franklin, and Perseverance. The Sample Analysis at Mars (SAM) instrument suite on Curiosity can detect a wide range of biosignatures, including volatile organics, polar and non-polar organics, complex organics (> 20 C atoms/molecule) and refractory organics 96 . SAM is a suite of three instruments that analyze volatiles released from rock or soil samples using three experimental methods: (1) evolved gas analysis-quadrupole mass spectrometry (EGA-QMS), in which samples are heated to ~ 800 °C and evolved gases are measured by QMS and GC-MS; (2) combustion, in which samples are heated in the presence of O 2 and the products are measured by QMS, GC-MS and/or tunable laser spectrometry (TLS); and (3) wet chemistry experiments, in which samples are derivatized at lower temperatures (up to ~ 300 °C) 24,96 . Onboard the Rosalind Franklin rover, biomarkers may be detected with a mass spectrometer, the Mars Organic Molecule Analyzer (MOMA), and a Raman spectrometer. MOMA is able to operate in two modes: pyrolysis/gas chromatography mass spectrometry (pyr/GC-MS) or laser desorption/ionization mass spectrometry (LDI-MS) at ambient Mars pressures 97,98 . The Mars2020 Perseverance rover, on the other hand, will carry a UV Raman spectrometer, the Scanning Habitable Environments with Raman and Luminescence for Organics and Chemicals (SHERLOC), using fine-scale imaging and a UV laser for the search of biomarkers 99 . Lipid biomarkers (alkanes, isoprenoids, fatty acids or alcohol series) detected with GC-MS techniques in Yungay clay-rich samples and relevant to the search for (a priori prokaryotic) Martian life, are in principle identifiable with the SAM, MOMA and SHERLOC instruments. In fact, lipid detection by SHERLOC should be used as a criterion for sample collection by Perseverance. Of the four fundamental biomolecules (lipids, nucleic acids, proteins, and carbohydrates), lipids have the greatest preservation potential over geologic time 100 . Carbonaceous chondrites could have carried abiogenic organic compounds to the Martian surface 101 , which can complicate the detection of biomarkers on Mars. The detection of an even-over-odd preference for fatty acid methyl esters (FAMEs, the building blocks of lipids), however, would be a strong indicator of biological processes 101,102 , assuming Martian biochemistry functions the same as terrestrial biochemistry. Still, key questions arise concerning relevant aspects such as matrix effects or limits of detection. For the GC-MS technique used in this study, samples require specific pre-treatment after organic solvent extraction and prior to analysis, such as separation of the total lipid extract into different polarity fractions (e.g., acidic, apolar and polar) or derivatization with trans-esterification (e.g., BF3) and tri-methylsilylation (e.g., BSTFA) reagents to convert the acidic and alcohol moieties into GC-MS amenable (i.e., volatile enough) compounds (FAMEs and trimethylsilyl ethers, respectively). SAM can perform wet chemistry experiments with tetramethylammonium hydroxide (TMAH) to make lipids sufficiently volatile to detect with GC-MS 103 and with N-methyl-N(tertbutyldimethylsilyl) trifluoroacetamide (MTBSTFA) to detect amino and carboxylic acids 96 . In fact, the first wet chemistry experiment was recently performed on active eolian Martian sand and results are being interpreted 104 . Curiosity is currently in the smectite-rich Glen Torridon unit, and SAM will perform wet chemistry experiments on a smectite-rich sample to search for complex organics 104 . In contrast, analyses with the MOMA and SHERLOC instruments will not involve such a complex pre-treatment. MOMA has the possibility of a derivatization step prior to GC-MS analysis, however, the lack of a previous fractionation step would affect the sample as a whole, not as polarity-specific moieties. As a spectroscopy technique, SHERLOC does not require any derivatization; however, based on fluorescence criteria, it is a technique susceptible to matrix effects that could complicate the detection of organics. Instrument sensitivity is also a key criterion for biomarker detection. SAM has a detection limit similar to the GC-MS technique used to analyze Yungay clay-rich samples (1 ppb or ng g −1 ). The detection limits for both the MOMA (1 nmol, which could vary between 100-600 ng depending on the molecular weight of the analyzed molecule) 97,98 and SHERLOC (in the range of ppm) instruments 99 , however, are orders of magnitude greater than SAM. Thus, SAM wet chemistry experiments in the smectite-rich Glen Torridon unit may be our best opportunity to identify biomarkers on Mars prior to sample return. The amount, concentration and distribution of target biomarkers (ppb or ppm) are critical for the success of the two near-future life-detection Martian missions.
A final point to consider for the detection of ancient Martian biomarkers is the alteration of organic molecules over time. The alteration and degradation of organic matter on Earth from burial diagenesis and thermal processes as they relate to hydrocarbon production have been well-studied for decades. Organic matter decomposes during early burial diagenesis typically via oxidation to CO 2 by microorganisms 105 . With progressive burial, heat and metamorphism, organic matter is first carbonized, removing most noncarbon elements and forming an aromatic skeleton, then graphitized through polymerization and structural rearrangement of the aromatic skeleton 106 . The oldest known sedimentary sequences on Earth (> 3.7 Gyr old) have been metamorphosed such that organic carbon has been graphitized, and whether the graphitized carbon represents traces of early life has been debated 107,108 . Mars, unlike Earth, never had robust plate tectonics 109 and, as a result, 3-4 Gyr old sedimentary rocks on the Martian surface have, for the most part, been minimally altered since their deposition. Even the ancient smectite-bearing mudstone in Gale crater lacks substantial mineralogical evidence for widespread burial diagenesis (i.e., there is no evidence for illitization) 110 despite a potential burial depth of up to 5 km 111 . Therefore, ancient smectite-bearing martian surfaces are ideal locations to investigate habitability in our early Solar System. Ancient sedimentary rocks on Mars have experienced some degree of diagenesis because they are Scientific Reports | (2020) 10:19183 | https://doi.org/10.1038/s41598-020-76302-z www.nature.com/scientificreports/ lithified, cemented and contain mineralized veins and concretions [112][113][114] . Diagenesis, in some cases, might work in favor of preserving organic matter on Mars. Thiophenic, aromatic and aliphatic compounds discovered by SAM in Gale crater have sulfur in their structures, suggesting early diagenesis in the presence of reduced sulfur made these molecules recalcitrant 25 . If sulfurization was a widespread process on early Mars, ancient organic molecules may have been preserved across the planet and may be indeed present in Jezero crater and Oxia Planum. Cosmic radiation on the Martian surface would also have a significant effect on biomarkers, especially those in the top 1 m of the surface 115 . Radiation oxidizes hydrocarbons and aromatic macromolecules to organic salts and CO 2 . The detection of organic molecules by SAM in multiple locations in Gale crater and the presence of organic molecules in carbonaceous chondrites, however, indicates that organic matter can survive long-term exposure to ionizing radiation 115 .

Conclusions
The results presented here, showing that wet subsurface clay minerals are inhabited by a number of metabolically active microorganisms in the midst of the driest place on earth, isolated and protected just centimeters bellow the extremely harsh surface environmental conditions typical of the Atacama, reinforce the notion that early Mars could have been a planet with similar subsurface protected habitable niches, particularly during the first billion years of its history. Orbital and in-situ measurements demonstrate a heterogeneous distribution of smectite and other clay minerals in Noachian-and early Hesperian-aged (~ 3.5 Gyr ago) surfaces 116,117 . Computer simulations and geochemical models suggest that clay minerals could have formed during relatively short warmer and wetter cycles in a predominantly dry and/or cold early Mars 118,119 . Thus, much of early Mars' history may have been similar to the modern Atacama, where wetting events were very rare. Because of colder conditions on early Mars, liquid water may have been generated by melting ice, rather than rain. The known patchy distribution of subsurface habitats in the hyperarid core of the Atacama 29,120,121 suggests that the search for biosignatures on Mars with the scientific payloads of future rovers will be a difficult task, as potential biosignatures could be similarly unevenly distributed throughout the Martian subsurface. Because the subsurface microbial community discovered in Yungay has adapted to an environment of persistent superficial drought with rare exposure to water, this expands potentially habitable environments on Mars to those with evidence for sporadic exposure to water, rather than persistent exposure. Fracture-associated halos in Gale crater, for example, extend into clay-rich rocks and represent locations where multiple fluid episodes altered the rock 122,123 . These clay-rich, subsurface zones may have been habitable to microorganisms similar to those found in subsurface soils in Yungay. If similar zones of aqueous alteration along fractures are discovered in Jezero crater, we suggest these should be high-priority targets for sample selection with Perseverance for eventual return to Earth. Our results also have implications for planetary protection constraints 124 , as ExoMars has a 2 m drill that could easily reach undetected, potentially inhabited clay minerals below the Martian surface. Ancient, smectitebearing rocks, however, may not be as habitable as unlithified soils because there may not be sufficient pore space for microorganisms to inhabit and the structural water in smectite-bearing rock may not be as readily available to microorganisms. The findings reported here may be considered as a guide to the search for hotspots preserving molecular biomarkers, and potentially life, well-protected below the Martian surface. Light penetration in the soil. Light penetration in the soil was measured in the field with a digital lux meter (Bestone Industrial Co., Ltd., Shenzhen, China). Measurements were taken by leaving the digital lux meter at the soil surface for 1 min, and then burying the sensor portion of the digital lux meter at one, and three centimeters of depth for the same amount of time.

Materials and methods
Environmental characterization (temperature and relative humidity). Temperature and relative humidity were measured in the field using dual iButton temperature/Humidity micro loggers (Maxim Integrated, San Jose, CA, USA) as previously done 46 , set to take data every 10 min during 28 July and 19 August 2018, and every 5 min between 22 and 29 July 2019. These sensors were set at the soil surface (shaded under a small rock) and also completely buried at 20 and 40 cm depth, with the sensor set at 40 cm well-immersed in the wet clay-rich soil. Environmental data statistical analyses: Pearson correlation coefficient. The number of degrees of freedom for r is n-2, where n is the number of pairs of bivariate data, for this case 782 for each correlation. Level of significance: alpha = 0.05. X-ray diffraction. Bulk samples were stored at room temperature and then ground in the lab into powder with an agate mortar and pestle (Pulverisette 2, Fristsch, Idar-Oberstein, Germany), and X-ray powder diffraction data were collected using a Bruker D8 Eco Advance (Massachusetts, USA) in Bragg-Brentano geometry, Cu Kα radiation and Lynxeye XE-T linear detector. The X-ray generator was operated at 40 kV and 25 mA. Samples were scanned with a 0.05° (2θ) step size, over the range 5-60° (2θ) with a 1 s collection time at each step. Phase identification was performed by comparing the measured diffraction pattern (diffractograms) with patterns of Scientific Reports | (2020) 10:19183 | https://doi.org/10.1038/s41598-020-76302-z www.nature.com/scientificreports/ the PDF Database with the DIFFRAC.EVA software (Bruker AXS, Massachusetts, USA). Oriented clay mounts were prepared using the filter-peel method 125 .
Clay characterization. Bulk samples in the lab were stored at room temperature and then dispersed in sodium hexametaphosphate (HMP 5%wt). Size fractionation was performed by low-speed centrifugation to obtain a suspension of the < 2 μm size fraction according to Stokes' law (also used for clay percentage determination). The suspension of clay-sized particles was titrated (acetate buffer, pH 5.0, until a pH of 6.8 was obtained) and washed with distilled water to remove carbonates and soluble salts, respectively.
Near-infrared spectroscopy. Near-infrared spectra of the clay-sized separates were stored at room temperature and then analyzed in the lab using the diffuse reflection method 126 (DRIFT) with a Nicolet FTIR spectrometer (Thermo Fisher Scientific, Massachusetts, USA). Samples were poured into a sample cup without dilution in KBr. Analysis were done at room temperature and under air-dried atmosphere. We used a mirror for background measurement. Spectra were collected using a DTGS-KBr detector at 2 cm −1 resolution in the NIR region (from 12,000 to 4000 cm −1 ; 1 to 2.5 μm) with a quartz beamsplitter.
Bacterial counts. Microorganisms in clay bearing soils were stored at room temperature and then enumer- . The finally obtained amplicons were validated and quantified by Bioanalyzer and an equimolecular pool was purified using AMPure beads (Beckman Coulter, Pasadena, USA) and titrated by quantitative PCR using the Kapa-SYBR FAST qPCR kit for LightCy-cler480 (Kapa Biosystems, Cape Town, South Africa) and a reference standard for quantification. The pool of amplicons was denatured prior to be seeded on a flowcell at a density of 10 pM, where clusters were formed and sequenced using a MiSeq Reagent Nano Kit v2 (Illumina, Berlin, Germany) in a 2 × 250 pair-end sequencing run on a MiSeq sequencer (Illumina, Berlin, Germany).
Raw sequences were processed in MOTHUR software v.1.40.5 128 , using a custom script based on MiSeq SOP 129 . Briefly, reads below 400 bp, with ambiguous nucleotide identities and/or homopolymers longer than 8 bp, singletons and putative chimeras were removed from subsequent analyses. Remaining reads were then clustered into OTUs (Operational Taxonomic Units) at the 97% similarity level. Number of sequences were finally normalized among samples to the lesser number of reads, i.e. 9823, by random selection. Taxonomic assignations were performed by comparing OTU's representative sequences with RDP database 130 , as it is the most up-to-date database that can be included in MOTHUR analyses, as well as the most accurate in assigning real taxonomic affiliation of OTUs obtained from desert samples 79 . OTUs assigned to 'cyanobacteria/chloroplast' were further compared with NBCI GenBank, EMBL, Greengenes and SILVA databases for more precise cyanobacteria taxonomic identification.  Korea). Sequences were checked for quality using the BioEdit software (Ibis Therapeutics, Carlsbad, USA) and end-trimmed before using the Megablast option for highly similar sequences of the BLASTN algorithm against the National Centre for Biotechnology Information nonredundant database (www.ncbi.nlm.nih.gov) to search for the closest species of each of the isolates obtained. Only species with at least 98% of sequence identify and an E value of 0.0 were selected, and only species with defined genus and species names were considered for phylogenetic closeness.
Phylogenetic analysis. In the lab, phylogenetic analysis of 16S rRNA gene sequences obtained from isolates were aligned by multiple sequence comparison by log-expectation 131 , analyzed with jModelTest 132 and then by Phylip NJ (bootstrap 10,000) 133 , all tools of the freely available Bosque phylogenetic analysis software (version 1.7.152) 134 .

Fluorescence microscopy, SYBR green and CTC staining. Small fragments of moist clay aggregates
from Yungay (about 1 cm 3 ) were carefully cut with a sterile blade in the lab. Plane clay surfaces were stained during 30 min at 37 °C in dark with 100 μL (1:100 dilution) of SYBR Green I (SBI) (Molecular Probes, Eugene, USA), a fluorochrome specific for staining nucleic acids. Multichannel Image Acquisition (MIA) system was used with combination of following filter sets: filter set for eGFP (Zeiss Filter Set 38; Ex ⁄ Em: 450-490 ⁄ 500-550 nm) and Rhodamine (Zeiss Filter Set 20; Ex ⁄ Em: 540-552 ⁄ 567-647 nm) and autofluorescence signal potentially proceeded from extracellular polymeric substances (EPS), respectively. The eGFP and DAPI (Zeiss Filter Set 49; Ex ⁄Em: 365 ⁄ 420-470 nm) filter sets were used for acquisition of single section images of SBI stained nucleic acids and autofluorescence signal potentially proceeded from EPS, respectively. Likewise, plane clay surfaces were stained during 24 h at 25 °C in dark with 100 μL of 5 mM of tetrazolium salt: 5-cyano-2,3-ditolyl tetrazolium chloride (CTC) (Molecular Probes, Eugene, USA) 135 . The formazan crystals are viewed as yellow-orange fluorescent spots when using Ex ⁄ Em: 426-446 ⁄ 545-645 nm filter set.
After both stains, a drop of immersion oil was placed on plane clay surfaces. After covering a clay surface with a square coverslip the samples were visualized with Zeiss AxioImager M.2 fluorescence microscope (Carl Zeiss, Jena, Germany) and a Plan-Apo 60x ⁄ 1.4 Zeiss oil-immersion objective. In some cases for better 3D visualization of bacteria aggregates, the Maximum Image Projection (MIP) images were reconstructed using the software package AxioVision 4.8.1 (Carl Zeiss, Jena, Germany).
EPS are a complex mixture of biomolecules; polysaccharides, proteins, nucleic acids, lipids and other macromolecules. Proteins and exopolysaccharides represent the key components of EPS, accounting for 40-95% of its structure. There are at least two components of EPS which can produce a signal in fluorescence microscopy; proteins and nucleic acids. We used SYBR Green staining for staining bacterial nucleic acids, thus disperse nucleic acids in the EPS should also be stained, leading to a detectable, although weaker, signal in FM. Some clays can produce a very weak autofluorescence signal, however, we did not observe any clay autofluorescence, but only the autofluorescence of proteins and the fluorescence coming from the SYBR Green stained nucleic acids.
Biosignature analyses. Biosignatures analysis were performed as previously reported by us 136 .
A. Geolipid extraction, fractionation and analysis Sediment sample (~ 50 g) was extracted with a mixture of dichloromethane/methanol (DCM/MeOH, 3:1, v/v) for 24 h with a Soxhlet apparatus (Fisher Scientific, New Hampshire, USA). Internal standards (tetracosane-D50, myristic acid-D27, 2-hexadecanol) were added prior to extraction. The total lipid extracts were concentrated by rotary evaporation to 2 ml. After this step, activated copper was added and allowed to stand overnight to remove elemental sulfur. The extracted sample was separated in two fractions using a Bond-elute column chromatography (bond phase NH2, 500 mg, 40 µm particle size). The neutral lipid fraction was obtained by eluting with 15 ml DCM/2-propanol (2:1,v/v) and the acid fraction with 15 ml of acetic acid (2%) in diethyl ether. Further separation of the neutral lipid fraction was done with 0.5 g of alumina in a Pasteur pipet. The non-polar fraction was obtained by eluting 4.5 ml of hexane/DCM (9:1, v/v) and the polar fraction with 3 ml of DCM/methanol (1:1, v/v). The acid fraction was derivatized with BF3 in methanol and the polar fraction with N,O-bis (trimethylsilyl) trifuoroacetamide (BSTFA).
Scientific Reports | (2020) 10:19183 | https://doi.org/10.1038/s41598-020-76302-z www.nature.com/scientificreports/ B. Gas chromatography-mass spectrometry (GC-MS) analysis The sample (non-polar, acid, and polar fraction) was analyzed by gas chromatography mass spectrometry using a 6850 GC system coupled to a 5975 VL MSD with a triple axis detector (Agilent Technologies, Santa Clara, USA) operating with electron ionization at 70 eV and scanning from 50 to 650 m/z. The analytes were injected (1 μl) and separated on a HP-5MS column (30 m × 0.25 mm i.d. × 0.25 μm film thickness) using He as a carrier gas at 1.1 ml min −1 . For the non-polar fraction, the oven temperature was programmed from 50 to 130 °C at 20 °C min −1 and then to 300 °C at 6 °C min −1 (held 20 min). For the acid fraction the oven temperature was programmed from 70 to 130 °C at 20 °C min −1 and then to 300 °C at 10 °C min −1 (held 10 min). For the polar fraction the oven temperature program was the same as that for the acid fraction, but the oven was held for 15 min at 300 °C. The injector temperature was 290 °C, the transfer line was 300 °C, and the MS source was 240 °C. Compound identification was based on the comparison of mass spectra and/ or reference compounds, compounds were quantified with external calibration curves. External standards of n-alkanes (C10 to C40), fatty acids methyl esters, FAMEs (C8 to C24), alkanols (C10, C14, C18, C20), and branched isoprenoids (2,6,10-trimethyl-docosane, crocetane, pristane, phytane, squalane and squalene) were injected to obtain calibration curves. Recoveries of the internal standards averaged 69 ± 18%. C. Stable isotopes analysis: organic carbon and total nitrogen Total nitrogen (δ 15 N) and organic carbon (δ 13 C) isotopes were measure by USGS methods 137 . Briefly, sediment sample (2 g) was homogenized by grinding with a corundum mortar and pestle. Then HCl was added to 0.5 g of sample to remove carbonate, equilibrated for 24 h, and adjusted to neutral pH with ultrapure water, dried at oven (50 °C), and analyzed in the IRMS (MAT 253, Thermo Fisher Scientific, Massachusetts, USA). δ 13 C and δ 15 N values were reported in the standard per mil notation using three certified standards (USGS41, IAEA-600, and USGS40), analytical precision of 0.1‰. D. Compound specific isotope analysis Conditions for gas chromatography analysis of isotopic ratio of n-alkanes, were identical with standard GC-MS analysis for the polar fraction. Carbon isotopic composition of individual compounds were performed using an isotope ratio gas-chromatograph-mass spectrometry (GC-IRMS) system, a Trace GC 1310 ultra, coupled to a MAT 253 IRMS (Thermo Fisher Scientific, Massachusetts, USA). Conditions in the IRMS were as follows: electron ionization 100 eV, Faraday cup collectors m/z 44, 45 and 46, a CuO/NiO combustion interface maintained at 1000 °C. The samples were injected in splitless mode (inlet temperature 250 °C, helium as a carrier gas at constant flow of 1.1 ml min −1 ). The isotopic values of peaks produced in the combustion reactor of the chromatography separated compounds were calculated using CO 2 -spikes of known isotopic composition, introduced directly into the source three time at the beginning and at the end of every run. An n-alkane reference mixture with known isotopic composition (A6, Indiana University, USA) were run after every four samples to check accuracy of the isotopic ratio determined by the GC-IRMS.